frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

An anchoring layer between facts and AI memory

https://github.com/Anchorstate-Lab/GMR
1•Zongming•46m ago

Comments

Zongming•46m ago
I’m trying to build a reliability layer between facts and AI memory. I call it the anchoring layer.

Most AI memory today is based on things the AI summarizes as “experience.” The problem is that facts can change, and the memory can become wrong.

When a new task starts, the goal, memory, and current facts can conflict. The model may then produce something different from what we expect.

The idea of the anchoring layer is to keep memories tied to the facts they came from, and make it easier to detect when they are no longer valid.

The project is here:

https://github.com/Anchorstate-Lab/GMR

I’d love to hear feedback on the idea.

Trump Gives Green Light to U.S. Companies to Aim Hacks at Cybercriminals

https://www.nytimes.com/2026/08/13/us/politics/trump-private-companies-hacking-cybercriminals.html
1•whyage•1m ago•0 comments

Standard RAG returns funny neighbors. Multi-vector fixes the signal

https://github.com/insightitsGit/VectorPrism
1•parvaamin•2m ago•0 comments

What We Lost When Computers Became Easier to Use [video]

https://www.youtube.com/watch?v=XHeMsXDyw2A
1•pierres7•9m ago•1 comments

Google Has Most of My Email Because It Has All of Yours (2014)

https://mako.cc/copyrighteous/google-has-most-of-my-email-because-it-has-all-of-yours
2•downbad_•13m ago•0 comments

Company Knowledge in ChatGPT (Business, Enterprise, and Edu)

https://help.openai.com/en/articles/12628342-company-knowledge-in-chatgpt-business-enterprise-and...
1•ssuds•17m ago•0 comments

Dangerous things should be difficult to do by accident

http://fexpr.blogspot.com/2011/05/dangerous-things-should-be-difficult-to.html
3•b-man•20m ago•0 comments

How AI text watermarking works

https://declaude.org/watermarking/
1•padolsey•21m ago•0 comments

Show HN: YazSes – offline hold-to-talk voice dictation for Linux/macOS/Windows

https://github.com/MSKazemi/yazses
1•mskazemi•22m ago•0 comments

I tried the new ChatGPT Desktop App for Linux – but I'll stick to my browser

https://www.zdnet.com/article/openai-brings-the-chatgpt-desktop-app-to-linux/
1•CrankyBear•22m ago•0 comments

Does anyone run Postgres without PgBouncer?

https://brandur.org/fragments/postgres-without-pgbouncer
1•birdculture•25m ago•0 comments

Oral History of John Chowning (Computer History) [video]

https://www.youtube.com/watch?v=e1Xn3030IvM
1•Rochus•25m ago•0 comments

AI is learning clinical judgment by practicing on simulated patients

https://www.echohive.ai/ai-clinical-residency
1•echohive42•30m ago•0 comments

Science investigation casts doubt on genes hailed for warding off Alzheimer's

https://www.science.org/content/article/science-investigation-casts-doubt-genes-hailed-warding-al...
1•sbulaev•31m ago•0 comments

Move FreeBSD System Between ZFS Disks

https://vermaden.wordpress.com/2026/08/14/move-freebsd-between-zfs-disks/
1•vermaden•31m ago•0 comments

PBS loses 70 years of TV history after cloud storage vendor goes defunct

https://www.tomshardware.com/software/cloud-storage/nine-pbs-loses-access-to-70-years-of-data-aft...
14•doener•33m ago•1 comments

The Coldcard Hack: A Postmortem

https://praveenperera.com/blog/coldcard-mk3-weak-rng-wave1/
1•abrowne2•33m ago•0 comments

Inside North Korea's Operation to Conquer the American Job Market

https://www.wsj.com/business/media/inside-north-koreas-operation-to-conquer-the-american-job-mark...
1•bookofjoe•36m ago•1 comments

The Evolving Role of the Red Team in the Era of Agentic Security

https://blog.google/security/the-evolving-role-of-the-red-team-in-the-era-of-agentic-security/
1•shaunpud•43m ago•0 comments

An anchoring layer between facts and AI memory

https://github.com/Anchorstate-Lab/GMR
1•Zongming•46m ago•1 comments

Further Developments About Internal AI Models Hacking Things

https://thezvi.substack.com/p/further-developments-about-internal
1•gmays•48m ago•0 comments

Show HN: AI Slop Kitchen

https://play-kitchen.netlify.app/
1•admtal•49m ago•0 comments

Think of AI as a lens, not a person

https://aethermug.com/posts/gen-ai-is-a-lens-into-humanity
2•mrcgnc•52m ago•0 comments

Loyalty cards that live in Apple/Google Wallet, no app to install

https://www.fideliyapass.com/
2•zakox•56m ago•0 comments

The Trump administration's Most Dangerous Attack on Science Yet

https://www.realclearscience.com/articles/2026/08/11/the_trump_administrations_most_dangerous_att...
5•rbanffy•57m ago•0 comments

Seoul soars past Dubai as busiest international airport

https://www.ft.com/content/f3f3c945-6a75-4663-b8ec-d5251438c1c5
3•JumpCrisscross•58m ago•0 comments

Watch It Work

https://www.texturehq.com/blog/watch-it-work
2•victorquinn•59m ago•0 comments

Show HN: Making Autonomous Work Reviewable

https://markupbase.com/whitepapers/making-autonomous-work-reviewable/
2•jasondoyle•59m ago•0 comments

14,000 crypto holders face security risk after Trezor data breach

https://www.ft.com/content/a91356ef-67bd-4bd9-947b-b272423f1318
3•petethomas•1h ago•0 comments

HoneyMatcha – A coordination platform for you and your personal agent

https://honeymatcha.io/
1•JaiRathore•1h ago•0 comments

Using AI chatbots? Here are four ways to reduce the energy drain

https://knowablemagazine.org/content/article/food-environment/2026/four-ways-to-reduce-energy-use...
1•knowablemag•1h ago•0 comments